There were no comments about fear hierarchy, fear level, occasional reinforced extinction, removal of safety signals, variability, retrieval cues, multiple contexts.
“The fear level of the participant is being monitored throughout the exposure trial.”
This could include affect labelling, but a crucial condition is that fear monitoring always involves monitoring the level of the emotion. We decide that this is sufficiently clear from the definition as we formulated it now; this should exclude affect labelling exercises where monitoring is not explicit.
“Expectancies for aversive events are explicitly stated before the start of the exposure trial.”
Regarding the first point, this is a methodological point; if indeed, in many therapy sessions/control groups, expectancies will be activated as ‘usual care’, the effect size will be very low. However, this would then accurately estimate the effects than can be expected of this therapy ingredient. We decide to not make any revisions based on this comment.
Regarding the second point, we decide to add monitoring as another therapy ingredient in addition to baseline statement of the expectancies. As formulation, we will use “Expectancies for aversive events are monitored throughout the exposure trial.”
“Following the exposure trial, the learning is consolidated by asking participants to judge what they learned regarding the non-occurrence of the feared event, discrepancies between what was predicted and what occurred.”
We decide to substitute ‘evaluation’ for consolidation. For clarity, we also rephrase the definition to: “Following the exposure trial, the participants are asked to evaluate what they learned regarding the non-occurrence of the feared event, discrepancies between what was predicted and what occurred.”
“The end of the exposure trial is determined by expectancy reduction to a certain level.”
In discussing this point, we realise that in fact, this therapy ingredient and ‘Fear level’ describe specific instances of generic ‘termination criteria’ that can be described as a combination of two decisions. First, does termination occur based on the monitored level of a psychological variable, or based on prespecified criteria that operationalise expectations regarding the influence of the exposure on the level of a psychological variable. Second, does this variable relate to somebody’s fear or their expectancies? The possible combinations would therefore be:
Note that we realise that not all combinations are plausible or realistic from the theoretical frameworks employed. However, the only cost of retaining them is that some may never be coded, which is a lower cost than omitting an therapy ingredient that does occur.
We will therefore also remove therapy ingredients ‘Expectancy violation’ and ‘Fear level’.
“Cognitive interventions designed to lessen probability overestimation (e.g., ???I am unlikely to be bitten by the dog???) and perceived negative valence (e.g., ???It is not so bad to be rejected???) occur.”
The first point is true; increasing self/response efficacy, coping skills, et cetera can also plausibly help reduce fear or change expectancies, and thus contribute to exposure therapy effectivess. However, such an intervention would not constitute exposure therapy per se, and therefore we decide to not code this element. (Note that we are aware that inclusion of a self-efficacy intervention might be a precondition for the effectiveness of some of these other therapy ingredients; but modeling such interactions goes far beyond the scope of what is feasible in this project.)
The second point is an excellent point, and directly concerns the goal of this review: whether this therapy ingredient is included in a therapy would be indicative of the theoretical perspective from which the therapy was designed.
The third point is also a good point; however, given the breadth of interventions applied to this end, we see no possibilities to specify this more explicitly. We will therefore, for now, retain this definition.
“Combination of multiple cues (internal and/or external) during exposure therapy, after initially conducting some exposure to each cue in isolation.”
This is a good point; we decide to add an additional therapy ingredient:
“The phobic stimulus is introduced for a brief period about 30 minutes before repeated trials of exposure.”
One participant remarked that ‘reconsilidation’ may be a misnomer; ‘retrieval-extinction’ may be more accurate. However, this participant did not think this was a redundant therapy ingredient to code. Consistent with our earlier decision to avoid ‘consolidation’, we will follow this suggestion and rename this therapy ingredient to “Brief pre-exposure”.
Indeed, number of treatments is important to code.
This is a good point. We will code this using the DSM categories.
Both points have been dealt with above.
We dealt with this above, as well.
This is an important omission. We will include this therapy ingredient:
We discussed this and decided that this is implicit in the ‘Nature of exposure’ entity that will be extracted. On the one hand, exposure should normally always result in fear elicitation. On the other hand, if we do want to extract/code this, it should be as a manipulation check of the variable ‘fear’; this is about therapy fidelity, not therapy ingredients.
Warning in readLines(con): line 26 appears to contain an embedded nul
NA
NA
NA
NA
NA
NA
Could also include ‘affect labelling’ more generally.
NA
NA
NA
NA
NA
NA
NA
NA
NA
I agree this is important, the only thing would be whether it can be shown to be an active ingredient - even without explicitly stating and disconfirming it, expectancies are likely to be activated and contradicted.
NA
NA
Expectancies for aversive events are explicitly stated before the start of the exposure trial and monitored throughout the exposure trial (perhaps better called expectancy monitoring)
NA
I would use another term than consolidation as this also refers to the process of memory storage/stabilisation that is also often spoken of in fear learning/extinction.
NA
NA
NA
NA
Could also be stated as ‘prediction error’. It’s a little difficult whether the expectancy violation should be ‘reduced to a certain level’, or whether it is instead a binary thing: you make the explicit and specific prediction and test it in the exposure. Could expectancy statement and expectancy violation even be combined to refer to a ‘behavioral test’ - as in CBT where the cognitions/expectations are treated as a hypothesis and the exposure functions as a behavioral test of them
NA
NA
NA
dit is 1 kant van angst; de andere is de mate waarin iemand zelfvertrouwen (self-efficacy; problem solving/coping) heeft een probleem aan te kunnen (CBT kan zich ook op die kant richten om angst te verminderen)
It would depend on when this occurs: if exposure therapy works by forcing a prediction error to update beliefs, then some theorists would say that lessening overestimation before exposure would be counterproductive, as it would lessen the contrast between what is expected and what occurs, and reduce inhibitory learning.
The formulation is somewhat obscure. What exactly is the intervention?
NA
NA
NA
Could also have a separate thing for enhancing extinction learning with pharmacological means? The same could go for habituation (a drug taken to reduce arousal during exposure could make habituation more rapid).
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
I would say ‘retrieval-extinction’ rather than reconsolidation. I the one case, you simply state what is occurring (brief retrieval, then extinction), whereas in the other a specific process is inferred. There is a lot of conflicting evidence about whether retrieval-extinction procedures are really reconsolidation-based.
NA
NA
NA
En waarschijnlijk ook goed om onderscheid te maken in typen angst: bijvoorbeeld in termen van complexiteit (bijv. spinnenangst vs. sociale angst).
The reference to reconsolidation is questionable as I noted above, in that we do not know whether reconsolidation is what is occurring. Also though, if a reconsolidation approach is taken and reconsolidation is what is believed to be happening, then it would be neither a habituation nor an inhibitory learning approach. The idea that people focusing on reconsolidation in exposure are pushing is that rather than inhibitory memories being formed, the original memory is directly updated.
No
contingency management (rewarding courageous behaviors (by praise or material rewards by therapist and carers; self-reward; ignoring anxious behaviors)
It’s maybe implicitly stated in the ‘fear monitoring’ ingredient, but I would add ‘fear eliciting’: one needs to be sure the participant is feeling anxious/experiencing fear in order for habituation to be possible
Assistant Professor in psychology
Prefer not to disclose
Full Professor in psychology
Prefer not to disclose
NA
NA
NA
NA
NA
both post doc and therapist
The first query is:
In this plot, the ‘AND’ operator is visualised by a solid line, while the ‘OR’ operator is visualised by a dotted line.
In this query, the searched terms must occur in entries’ title, abstract fields.
In the interface language of the PubMed interface to the PubMed database and the Ebscohost and Ovid interfaces to a variety of databases such as PsycINFO, PsycArticles, and MedLine, this query renders as:
PUBMED QUERY:
(((cognitive behavio* therapy [TIAB]) OR (cbt [TIAB])) AND ((exposure [TIAB])) AND ((child* [TIAB]) OR (adolescent* [TIAB]) OR (youth [TIAB])) AND ((anxiety [TIAB]) OR (fear [TIAB])))
EBSCOHOST QUERY:
((TI ("cognitive behavio* therapy" OR "cbt")) OR (AB ("cognitive behavio* therapy" OR "cbt"))) AND ((TI ("exposure")) OR (AB ("exposure"))) AND ((TI ("child*" OR "adolescent*" OR "youth")) OR (AB ("child*" OR "adolescent*" OR "youth"))) AND ((TI ("anxiety" OR "fear")) OR (AB ("anxiety" OR "fear")))
OVID QUERY:
((("cognitive behavio* therapy" OR "cbt") AND ("exposure") AND ("child*" OR "adolescent*" OR "youth") AND ("anxiety" OR "fear"))).ti,ab
NOTE: export the results as .RIS files, called 'MEDLINE' in PubMed.
This query was run at 2018-08-28 in PubMed (XXXX hits; file saved as pubmed-2018-08-28.ris) and PsycINFO accessed through EbscoHost (XXXX hits; file saved as psycinfo-2018-08-28.ris), and exported to RIS format (called ‘MEDLINE’ in PubMed). The RIS files were then imported in R using metabefor.
### Import PsycINFO hits
firstQueryIteration_psycinfo <-
importRISlike(file.path(queryHitExportPath,
queries[[1]]$date,
paste0(queries[[1]]$date, "--",
queries[[1]]$databases[[1]]$interface, "--",
names(queries[[1]]$databases)[1], ".",
queries[[1]]$databases[[1]]$fileFormat)),
encoding="native.enc");
Reading file 'B:/Data/research/habituation-versus-inhibition-in-exposure-therapy/queries/2018-08-28/2018-08-28--ebsco--psycinfo.ris'...
Read 8555 lines.
Extracted 205 lines matching regex '^TY' (regular RIS start of record).
Extracting references...
Interpreting references...
Converting references to dataframe...
### Import PubMed hits
firstQueryIteration_pubmed <-
importRISlike(file.path(queryHitExportPath,
queries[[1]]$date,
paste0(queries[[1]]$date, "--",
queries[[1]]$databases[[2]]$interface, "--",
names(queries[[1]]$databases)[2], ".",
queries[[1]]$databases[[2]]$fileFormat)),
encoding="native.enc");
Reading file 'B:/Data/research/habituation-versus-inhibition-in-exposure-therapy/queries/2018-08-28/2018-08-28--pubmed--pubmed.ris'...
Read 10946 lines.
Extracted 0 lines matching regex '^TY' (regular RIS start of record).
Zero hits: looked for PubMed RIS export ('medline') markers:
Extracted 126 lines matching regex '^PMID' (PubMed RIS start of record).
Extracting references...
Interpreting references...
Converting references to dataframe...
### Merge the two sets of hits
firstQueryIteration <-
findDuplicateReferences(primaryRefs = firstQueryIteration_psycinfo,
secondaryRefs = firstQueryIteration_pubmed,
duplicateFieldValue = "dupl",
newRecordValue = "PubMed",
duplicateValue = "duplicate (both PsycINFO and PubMed)",
originalValue = "PsycINFO");
Merging bibliographic databases and flagging duplicates. Processing 205 primary references and 126 secondary references. It is now 2018-11-19 10:24:53.
Processed 205 primary and 126 secondary records. Identified 1 duplicate records. Total number of records in resulting set is 331, of which 1 duplicates. It is now 2018-11-19 10:25:01. The process took roughly: Time difference of 8.110758 secs.
### Generate bibtex keys
firstQueryIteration$output$records <-
generateBibtexkeys(firstQueryIteration$output$records);
### Add query date identifier to bibtex keys
firstQueryIteration$output$records$bibtexkey <-
paste0(firstQueryIteration$output$records$bibtexkey,
"-", gsub("-", "", queries[[1]]$date));
screening1_filename_pre <- paste0(queries[[1]]$date, "-screening.bib");
screening1_filename_post <- paste0(queries[[1]]$date, "-screened.bib");
### Export the hits to bibtex for screening in JabRef
sysrevExport(firstQueryIteration,
filename=file.path(screeningPath,
screening1_filename_pre),
screeningType=NULL);
The merged list of query hits has now been exported to file 2018-08-28-screening.bib in directory “screening” and can be opened using JabRef, which can be downloaded from https://www.fosshub.com/JabRef.html.
When opening a bibliographic library (i.e. a file with the extension .bib) in JabRef, it will show the entry table, which is a convenient way to inspect all entries (hits, references, articles, etc) in the library. To prepare JabRef for screening, two settings are important.
First, to change the fields that are visible in the overview table of all references (i.e. the entry table), open the ‘Options’ drop-down menu and select ‘Preferences’. In the preferences dialog, open the ‘Entry table columns’ section:
Figure 1: Screenshot of JabRef preferences dialog when the ‘Entry table columns’ section is opened.
There, the columns shown in the entry table can be edited in the ‘Entry table columns’ sections. A bit confusingly, this is done by adding rows in the table shown in this dialog. Each ‘row’ in this table represents a column in the entry table.
Note that in bibtex (and therefore JabRef), you can create new fields on the fly. In this case, use field ‘screening1’ for screening the hits of this first screening iteration: simply add this field name as a ‘row’ (column) in the entry table. This will show, for every entry, the contents of that field (if it has any).
Second, you need to be able to edit the content in that field. The entry table is very convenient to maintain an overview of the entries in the database, but cannot be used for editing. To edit an entry, double click it in the entry tabel. This opens the entry editor, which has a number of tabs. Each tab shows a number of fields which can then be edited.
These tabs can be configured by setting the ‘General fields’. Open the ‘Options’ drop-down menu and select ‘General Fields’ to configure which fields are available in the different tabs when opening an entry.
Figure 2: Screenshot of JabRef dialog used to set the general fields.
Add a dedicated field for the reviewing, showing only the title, abstract, and screening1 fields. This allows you to focus on the relevant information while ignoring irrelevant and potentially biasing information (such as year, journal, and authors). Each row in this text area shows one tab. The first term on each row is the tab’s name, followed by a colon (:) and then the fields shown in the tab, separated by semicolons (;). For example, you could add the following row:
Screening Round 1:title;abstract;screening1
For every entry, add the following text in the ‘screening’ field:
dupl if the study is a duplicate of another entry;noengl if the study is not reported in English;noexper if the study does not have an experimental design;nopopul if the study did not sample participants younger than 18 years;noexpos if the study did not compare two groups that differ in the treatment in terms of exposure as a part of cognitive behavioral therapy;nophobia if the study did not concern treatment for phobia disorders;incl.So once JabRef is opened, when screening, make sure that the ‘screening1’ field is shown in the entry table (i.e. that it is one of the entry table columns), and create one entry editing tab using ‘General Fields’ that contains the fields title, abstract, and screening1. You can then use this tab for the screening. It is also convenient to show field dupl in either the entry table or the screening tab in the entry editor, because for duplicate records (that were identified as such - the algorithm may miss some duplicates of course), that field contains the text dupl.
Make sure to save the database with query hits under a different name than 2018-08-28-screening.bib. That is important because file 2018-08-28-screening.bib will get overwritten if this R Markdown file is executed again. This file will not require any adjustments if you name the database 2018-08-28-screened.bib.
This is an overview of the screening results. The details for the sources to include are listed in the second tab.
Converting references to dataframe...
Frequencies Perc.Total Perc.Valid Cumulative
dupl 68 20.5 20.5 20.5
incl 13 3.9 3.9 24.5
noengl 1 0.3 0.3 24.8
noexper 230 69.5 69.5 94.3
noexpos 15 4.5 4.5 98.8
nophobia 2 0.6 0.6 99.4
nopopul 2 0.6 0.6 100.0
Total valid 331 100.0 100.0
Nixon, Reginald D. V. and Sterk, Jisca and Pearce, Amanda and Weber, Nathan (2017/07//) A randomized trial of cognitive behavior therapy and cognitive therapy for children with posttraumatic stress disorder following single-incident trauma: Predictors and outcome at 1-year follow-up.. Psychological Trauma: Theory, Research, Practice, and Policy. 10.1037/tra0000190
Objective: The 1-year outcome and moderators of adjustment for children and youth receiving treatment for posttraumatic stress disorder (PTSD) following single-incident trauma was examined. Method: Children and youth who had experienced single-incident trauma (N = 33; 7–17 years old) were randomly assigned to receive 9 weeks of either trauma-focused cognitive behavior therapy (CBT) or trauma-focused cognitive therapy (without exposure; CT) that was administered to them and their parents individually. Results: Intent-to-treat analyses demonstrated that both groups maintained posttreatment gains in PTSD, depression and general anxiety symptoms reductions at 1-year follow-up, with no children meeting criteria for PTSD. A large proportion of children showed good end-state functioning at follow-up (CBT: 65%; CT: 71%). Contrary to 6-month outcomes, maternal adjustment no longer moderated children’s outcome, nor did any other tested variables. Conclusion: The findings confirm the positive longer-term outcomes of using trauma-focused cognitive–behavioral methods for PTSD secondary to single-incident trauma and that these outcomes are not dependent on the use of exposure. (PsycINFO Database Record (c) 2017 APA, all rights reserved)
Wood, Jeffrey J. and Ehrenreich-May, Jill and Alessandri, Michael and Fujii, Cori and Renno, Patricia and Laugeson, Elizabeth and Piacentini, John C. and De Nadai, Alessandro S. and Arnold, Elysse and Lewin, Adam B. and Murphy, Tanya K. and Storch, Eric A. (2015/01//) Cognitive behavioral therapy for early adolescents with autism spectrum disorders and clinical anxiety: A randomized, controlled trial.. Behavior Therapy. 10.1016/j.beth.2014.01.002
Clinically elevated anxiety is a common, impairing feature of autism spectrum disorders (ASD). A modular CBT program designed for preteens with ASD, Behavioral Interventions for Anxiety in Children with Autism (BIACA; Wood et al., 2009) was enhanced and modified to address the developmental needs of early adolescents with ASD and clinical anxiety. Thirty-three adolescents (11–15years old) were randomly assigned to 16 sessions of CBT or an equivalent waitlist period. The CBT model emphasized exposure, challenging irrational beliefs, and behavioral supports provided by caregivers, as well as numerous ASD-specific treatment elements. Independent evaluators, parents, and adolescents rated symptom severity at baseline and posttreatment/postwaitlist. In intent-to-treat analyses, the CBT group outperformed the waitlist group on independent evaluators’ ratings of anxiety severity on the Pediatric Anxiety Rating Scale (PARS) and 79% of the CBT group met Clinical Global Impressions–Improvement scale criteria for positive treatment response at posttreatment, as compared to only 28.6% of the waitlist group. Group differences were not found for diagnostic remission or questionnaire measures of anxiety. However, parent-report data indicated that there was a positive treatment effect of CBT on autism symptom severity. The CBT manual under investigation, enhanced for early adolescents with ASD, yielded meaningful treatment effects on the primary outcome measure (PARS), although additional developmental modifications to the manual are likely warranted. Future studies examining this protocol relative to an active control are needed. (PsycINFO Database Record (c) 2016 APA, all rights reserved)
Sung, Min and Ooi, Yoon Phaik and Goh, Tze Jui and Pathy, Pavarthy and Fung, Daniel S. S. and Ang, Rebecca P. and Chua, Alina and Lam, Chee Meng (2011/12//) Effects of cognitive-behavioral therapy on anxiety in children with autism spectrum disorders: A randomized controlled trial.. Child Psychiatry and Human Development. 10.1007/s10578-011-0238-1
We compared the effects of a 16-week Cognitive-Behavioral Therapy (CBT) program and a Social Recreational (SR) program on anxiety in children with Autism Spectrum Disorders (ASD). Seventy children (9–16 years old) were randomly assigned to either of the programs (nCBT = 36; nSR = 34). Measures on child’s anxiety using the Spence Child Anxiety Scale—Child (SCAS-C) and the Clinical Global Impression—Severity scale (CGI-S) were administered at pre-, post-treatment, and follow-ups (3- and 6-month). Children in both programs showed significantly lower levels of generalized anxiety and total anxiety symptoms at 6-month follow-up on SCAS-C. Clinician ratings on the CGI-S demonstrated an increase in the percentage of participants rated as ‘‘Normal’’ and ‘‘Borderline’’ for both programs. Findings from the present study suggest factors such as regular sessions in a structured setting, consistent therapists, social exposure and the use of autism-friendly strategies are important components of an effective framework in the management of anxiety in children and adolescents with ASD. (PsycINFO Database Record (c) 2016 APA, all rights reserved)
Rudy, Brittany M. and Zavrou, Sophia and Johnco, Carly and Storch, Eric A. and Lewin, Adam B. (2017/09//) Parent-led exposure therapy: A pilot study of a brief behavioral treatment for anxiety in young children.. Journal of Child and Family Studies. 10.1007/s10826-017-0772-y
Despite prevalence rates as high as 9.4%, few studies have examined the applicability of cognitive-behavioral therapy for treatment of anxiety disorders in very young children (i.e., below the age of 7 years). The present study examined the preliminary efficacy of a parent-led exposure therapy protocol (PLET) designed for young children with anxiety disorders. Twenty-two youth aged 4–7 years and their parents participated in this pilot randomized control trial. Families of youth with significant anxiety concerns were randomized to either PLET (n = 12), a 10 session/5-week family-based exposure therapy program designed to target anxiety in young children, or treatment as usual (TAU; n = 10). Children in the PLET condition demonstrated a greater reduction in anxiety symptoms compared to TAU (d = 3.18), with 90.91% of PLET participants (active condition) being classified as treatment responders at post-treatment as opposed to 0 in the TAU condition. Gains were maintained at 1 month-follow-up. Although pilot in nature, these data suggest in a preliminary fashion that a parent led exposure therapy protocol that is adapted appropriately for developmental age and incorporates an active coaching component for parents may be efficacious for the treatment of young children with anxiety disorders. (PsycINFO Database Record (c) 2018 APA, all rights reserved)
Ginsburg, Golda S. and Drake, Kelly L. (2002/07//) School-based treatment for anxious African-American adolescents: A controlled pilot study.. Journal of the American Academy of Child & Adolescent Psychiatry. 10.1097/00004583-200207000-00007
Evaluated the feasibility and effectiveness of a school-based group cognitive-behavioral treatment (CBT) for anxiety disorders with African-American adolescents. Twelve adolescents (mean age = 15.6 years) with anxiety disorders were randomly assigned to CBT (n = 6) or a group attention-support control condition (AS-Control; n = 6). Both groups met for 10 sessions in the same high school. Key treatment ingredients in CBT involved exposure, relaxation, social skills, and cognitive restructuring. Key ingredients in AS-Control involved therapist and peer support. At preand posttreatment, diagnostic interviews were conducted, and adolescents completed self-report measures of anxiety. At posttreatment and among those who attended more than one treatment session, 3/4 adolescents in CBT no longer met diagnostic criteria for their primary anxiety disorder, compared with 1/5 in AS-Control. Clinician ratings of impairment and self-report levels of overall anxiety were significantly lower at posttreatment in CBT compared with ASControl. Teenagers in both groups reported lower levels of social anxiety from pre- to posttreatment. Findings support the feasibility of implementing a manual-based CBT in an urban school setting. (PsycINFO Database Record (c) 2016 APA, all rights reserved)
Whiteside, Stephen P. H. and Ale, Chelsea M. and Young, Brennan and Dammann, Julie E. and Tiede, Michael S. and Biggs, Bridget K. (2015/10//) The feasibility of improving CBT for childhood anxiety disorders through a dismantling study.. Behaviour Research and Therapy. 10.1016/j.brat.2015.07.011
This preliminary randomized controlled trial (RCT) examines the feasibility of dismantling cognitive behavioral therapy (CBT) for childhood anxiety disorders. Fourteen children (10 girls) ages 7 to 14 (m = 10.2) with social phobia, generalized anxiety disorder, separation anxiety disorder, or panic disorder were randomized to receive 6 sessions of either a) the pre-exposure anxiety management strategies presented in traditional CBT, or b) parent-coached exposure therapy. The sample was selected from a treatment seeking population and is representative of children in clinical settings. Examination of fidelity ratings, dropouts, and satisfaction ratings indicated that the interventions were distinguishable, safe, and tolerable. The overall sample improved significantly with pre-post effect sizes generally in the large range for both conditions. Between-group effect sizes indicating greater improvement with parent-coached exposure therapy were moderate or large for ten of 12 variables (i.e., 0.53 to 1.52). Re-evaluation after three months of open treatment suggested that the intervention emphasizing exposure early maintained its superiority while requiring fewer appointments. (PsycINFO Database Record (c) 2018 APA, all rights reserved)
Spence, Susan H. and Donovan, Caroline and Brechman-Toussaint, Margaret (2000/09//) The treatment of childhood social phobia: The effectiveness of a social skills training-based, cognitive-behavioural intervention, with and without parental involvement.. Journal of Child Psychology and Psychiatry. 10.1111/1469-7610.00659
50 children (aged 7–14 yrs) with a principal diagnosis of social phobia were randomly assigned to either child-focused cognitive-behavior therapy (CBT), CBT plus parent involvement, or a wait list control (WLC). The integrated CBT program involved intensive social skills training combined with graded exposure and cognitive challenging. At posttreatment, significantly fewer children in the treatment conditions retained a clinical diagnosis of social phobia compared to the WLC condition. In comparison to the WLC, children in both CBT interventions showed significantly greater reductions in children’s social and general anxiety and a significant increase in parental ratings of child social skills performance. At 12-mo follow-up, both treatment groups retained their improvement. There was a trend towards superior results when parents were involved in treatment, but this effect was not statistically significant. (PsycINFO Database Record (c) 2016 APA, all rights reserved)
Ginsburg, Golda S. and Becker, Kimberly D. and Drazdowski, Tess K. and Tein, Jenn-Yun (2012/02//) Treating anxiety disorders in inner city schools: Results from a pilot randomized controlled trial comparing CBT and usual care.. Child & Youth Care Forum. 10.1007/s10566-011-9156-4
Background: The effectiveness of cognitive-behavioral treatment (CBT) in inner city schools, when delivered by novice CBT clinicians, and compared to usual care (UC), is unknown. Objective: This pilot study addressed this issue by comparing a modular CBT for anxiety disorders to UC in a sample of 32 volunteer youth (mean age 10.28 years, 63% female, 84% African American) seen in school-based mental health programs. Methods: Youth were randomly assigned to CBT (n = 17) or UC (n = 15); independent evaluators conducted diagnostic interviews with children and parents at pre- and post-intervention, and at a one-month follow-up. Results: Based on intent-to-treat analyses, no differences were found in response rates between groups with 50 and 42% of the children in CBT, compared to 46 and 57% in UC no longer meeting criteria for an anxiety disorder at post-treatment and follow-up respectively. Similar improvements in global functioning were also found in both treatment groups. Baseline predictors of a positive treatment response included lower anxiety, fewer maladaptive thoughts, less exposure to urban hassles, and lower levels of parenting stress. Therapist use of more CBT session structure elements and greater competence in implementing these elements was also related to a positive treatment response. Conclusions: Findings from this small pilot failed to show that CBT was superior to UC when delivered by school-based clinicians. Large scale comparative effectiveness trials are needed to determine whether CBT leads to superior clinical outcomes prior to dissemination. (PsycINFO Database Record (c) 2016 APA, all rights reserved)
Chu BC and Crocco ST and Esseling P and Areizaga MJ and Lindner AM and Skriner LC (2016 Jan) Transdiagnostic group behavioral activation and exposure therapy for youth anxiety and depression: Initial randomized controlled trial.. Behaviour research and therapy. 10.1016/j.brat.2015.11.005 [doi], S0005-7967(15)30054-1 [pii]
Anxiety and depression are debilitating and commonly co-occurring in young adolescents, yet few interventions are designed to treat both disorder classes together. Initial efficacy is presented of a school-based transdiagnostic group behavioral activation therapy (GBAT) that emphasizes anti-avoidance in vivo exposure. Youth (N = 35; ages 12-14; 50.9% male) were randomly assigned to either GBAT (n = 21) or WL (n = 14) after completing a double-gated screening process. Multi-reporter, multi-domain outcomes were assessed at pretreatment, posttreatment, and four-month follow-up (FU). GBAT was associated with greater posttreatment remission rates than WL in principal diagnosis (57.1% vs. 28.6%; X1(2) = 2.76, p = .09) and secondary diagnosis (70.6% vs. 10%; X1(2) = 9.26, p = .003), and greater improvement in Clinical Global Impairment - Severity ratings, B = -1.10 (0.42), p = .01. Symptom outcomes were not significantly different at posttreatment. GBAT produced greater posttreatment behavioral activation (large effect size) and fewer negative thoughts (medium effect), two transdiagnostic processes, both at the trend level. Most outcomes showed linear improvement from pretreatment to FU that did not differ depending on initial condition assignment. Sample size was small, but GBAT is a promising transdiagnostic intervention for youth anxiety and unipolar mood disorders that can feasibly and acceptably be applied in school settings.
Peris TS and Compton SN and Kendall PC and Birmaher B and Sherrill J and March J and Gosch E and Ginsburg G and Rynn M and McCracken JT and Keeton CP and Sakolsky D and Suveg C and Aschenbrand S and Almirall D and Iyengar S and Walkup JT and Albano AM and Piacentini J (2015 Apr) Trajectories of change in youth anxiety during cognitive-behavior therapy.. Journal of consulting and clinical psychology. 10.1037/a0038402 [doi]
OBJECTIVE: To evaluate changes in the trajectory of youth anxiety following the introduction of specific cognitive-behavior therapy (CBT) components: relaxation training, cognitive restructuring, and exposure tasks. METHOD: Four hundred eighty-eight youths ages 7-17 years (50% female; 74% </= 12 years) were randomly assigned to receive either CBT, sertraline (SRT), their combination (COMB), or pill placebo (PBO) as part of their participation in the Child/Adolescent Anxiety Multimodal Study (CAMS). Youths in the CBT conditions were evaluated weekly by therapists using the Clinical Global Impression Scale-Severity (CGI-S; Guy, 1976) and the Children’s Global Assessment Scale (CGAS; Shaffer et al., 1983) and every 4 weeks by blind independent evaluators (IEs) using the Pediatric Anxiety Ratings Scale (PARS; RUPP Anxiety Study Group, 2002). Youths in SRT and PBO were included as controls. RESULTS: Longitudinal discontinuity analyses indicated that the introduction of both cognitive restructuring (e.g., changing self-talk) and exposure tasks significantly accelerated the rate of progress on measures of symptom severity and global functioning moving forward in treatment; the introduction of relaxation training had limited impact. Counter to expectations, no strategy altered the rate of progress in the specific domain of anxiety that it was intended to target (i.e., somatic symptoms, anxious self-talk, avoidance behavior). CONCLUSIONS: Findings support CBT theory and suggest that cognitive restructuring and exposure tasks each make substantial contributions to improvement in youth anxiety. Implications for future research are discussed. (PsycINFO Database Record
Leutgeb V and Schafer A and Kochel A and Schienle A (2012 Apr) Exposure therapy leads to enhanced late frontal positivity in 8- to 13-year-old spider phobic girls.. Biological psychology. 10.1016/j.biopsycho.2012.02.008 [doi]
Neurobiological studies have demonstrated that psychotherapy is able to alter brain function in adults, however little exists on this topic with respect to children. This waiting-list controlled investigation focused on therapy-related changes of the P300 and the late positive potential (LPP) in 8- to 13-year-old spider phobic girls. Thirty-two patients were presented with phobia-relevant, generally disgust-inducing, fear-inducing, and affectively neutral pictures while an electroencephalogram was recorded. Participants received one session of up to 4h of cognitive-behavioral exposure therapy. Treated children showed enhanced amplitudes of the LPP at frontal sites in response to spider pictures. This result is interpreted to reflect an improvement in controlled attentional engagement and is in line with already existing data for adult females. Moreover, the girls showed a therapy-specific reduction in overall disgust proneness, as well as in experienced arousal and disgust when viewing disgust pictures. Thus, exposure therapy seems to have broad effects in children.
Nixon RD and Sterk J and Pearce A (2012 Apr) A randomized trial of cognitive behaviour therapy and cognitive therapy for children with posttraumatic stress disorder following single-incident trauma.. Journal of abnormal child psychology. 10.1007/s10802-011-9566-7 [doi]
The present study compared the efficacy of trauma-focused cognitive behavior therapy (CBT) with trauma-focused cognitive therapy (without exposure; CT) for children and youth with posttraumatic stress disorder (PTSD). Children and youth who had experienced single-incident trauma (N = 33; 7-17 years old) were randomly assigned to receive 9 weeks of either CBT or CT which was administered individually to children and their parents. Intent-to-treat analyses demonstrated that both interventions significantly reduced severity of PTSD, depression, and general anxiety. At posttreatment 65% of CBT and 56% of the CT group no longer met criteria for PTSD. Treatment completers showed a better response (CBT: 91%; CT: 90%), and gains were maintained at 6-month follow-up. Maternal depressive symptoms and unhelpful trauma beliefs moderated children’s outcome. It is concluded that PTSD secondary to single-incident trauma can be successfully treated with trauma-focused cognitive behavioural methods and the use of exposure is not a prerequisite for good outcome.
Kendall PC and Flannery-Schroeder E and Panichelli-Mindel SM and Southam-Gerow M and Henin A and Warman M (1997 Jun) Therapy for youths with anxiety disorders: a second randomized clinical trial.. Journal of consulting and clinical psychology. 10.1037/0022-006X.65.3.366
Ninety-four children (aged 9-13 years) with anxiety disorders were randomly assigned to cognitive behavioral treatment or waiting-list control. Outcomes were evaluated using diagnostic status, child self-reports, parent and teacher reports, cognitive assessment and behavioral observation: maintenance was examined using 1-year follow-up data. Analyses of dependent measures indicated significant improvements over time, with the majority indicating greater gains for those receiving treatment. Treatment gains returned cases to within nondeviant limits (i.e., normative comparisons) and were maintained at 1-year follow-up. Client age and comorbid status did not moderate outcomes. A preliminary examination of treatment segments suggested that the enactive exposure (when it follows cognitive-educational training) was an active force in beneficial change. Discussion includes suggestions for future research.
### Import PsycINFO hits
secondQueryIteration_psycinfo <-
importRISlike(file.path(queryHitExportPath,
"psycinfo-2018-05-23.ris"),
encoding="native.enc");
### Import PubMed hits
secondQueryIteration_pubmed <-
importRISlike(file.path(queryHitExportPath,
"pubmed-2018-05-23.ris"));
### Merge the two sets of hits
secondQueryIteration <-
findDuplicateReferences(primaryRefs = secondQueryIteration_psycinfo,
secondaryRefs = secondQueryIteration_pubmed,
duplicateFieldValue = "dupl (2nd)",
newRecordValue = "PubMed (2nd)",
duplicateValue = "duplicate (both PsycINFO and PubMed; 2nd)",
originalValue = "PsycINFO (2nd)");
### Generate bibtex keys
secondQueryIteration$output$records <-
generateBibtexkeys(secondQueryIteration$output$records);
### Add query date identifier to bibtex keys
secondQueryIteration$output$records$bibtexkey <-
paste0(secondQueryIteration$output$records$bibtexkey,
"-20180523");
### Import results from first query (these have been screened now)
firstQueryIteration_screened <-
importBibtex(file.path(screeningPath,
"2018-05-14-screening#1.bib"));
### Merge the screened reference database from the first query
### with the database from the second query
secondQueryIteration_merged <-
findDuplicateReferences(primaryRefs = firstQueryIteration_screened,
secondaryRefs = secondQueryIteration,
duplicateFieldValue = "Screened in first iteration",
newRecordValue = "From second query",
duplicateValue = "From first query (screened in first iteration)",
originalValue = "screening1");
### The new records are stored in secondQueryIteration_merged$output$newRecords, so we
### can copy these to the database from the first screening. We also store the entire
### database so that we can document the process (and if need be, check whether anything
### went wrong).
secondScreening <- firstQueryIteration_screened;
secondScreening$output$records <- rbind.fill(secondScreening$output$records,
secondQueryIteration_merged$output$newRecords);
### Export the hits to bibtex for screening in JabRef
sysrevExport(secondQueryIteration_merged,
filename=file.path(screeningPath,
"2018-05-23-fully-merged-database.bib"),
screeningType="screening");
sysrevExport(secondScreening,
filename=file.path(screeningPath,
"2018-05-23-screening.bib"),
screeningType="screening");
We will use a metabefor extraction script for the extraction of the data. The idea of this script is to extract the data from the original sources with a minimum of interpretation. The data is extracted into a machine-readable format, which then allows competely transparent further processing and synthesis.
These scripts are generated on the basis of two tables/spreadsheets. The first contains the entities to extract, such as study year, sample size, how variables were operationalised, and associations that were found. The second contains the valid values for each entity, to allow efficiently providing coders with examples, instructions, and to allow easy verification of the input.
The logged messages from this process are available in this section under the tab ‘Logged messages’, and the generated extraction script template (which is also written as a file to the repository) is included in a text area in the ‘Extraction script template’ for convenient inspection.
Sheet-identifying info appears to be a browser URL.
googlesheets will attempt to extract sheet key from the URL.
Putative key: 1TpdpB926luKVy2tCwBusFkxZ7xv-BzdzzBjmOU26PI8
Worksheets feed constructed with public visibility
Accessing worksheet titled 'entities'.
Parsed with column specification:
cols(
title = col_character(),
description = col_character(),
identifier = col_character(),
valueTemplate = col_character(),
validValues = col_character(),
default = col_character(),
examples = col_character(),
parent = col_character(),
entityRef = col_character(),
fieldRef = col_character(),
owner = col_character(),
list = col_logical(),
collapsing = col_character(),
repeating = col_logical(),
recurring = col_character(),
recursing = col_character(),
identifying = col_logical()
)
Accessing worksheet titled 'valueTemplates'.
Parsed with column specification:
cols(
identifier = col_character(),
description = col_character(),
validValues = col_character(),
default = col_character(),
examples = col_character(),
validation = col_character(),
error = col_character()
)
Accessing worksheet titled 'definitions'.
Parsed with column specification:
cols(
Term = col_character(),
Definition = col_character()
)
Successfully read the extraction script specifications from Google sheets.
Stored local backup of entities to 'B:/Data/research/habituation-versus-inhibition-in-exposure-therapy/extraction/entities-local-copy.csv'.
Stored local backup of value templates to 'B:/Data/research/habituation-versus-inhibition-in-exposure-therapy/extraction/valueTemplates-local-copy.csv'.
Parsed extraction script specifications into extraction script template.
Successfully wrote extraction script template to 'B:/Data/research/habituation-versus-inhibition-in-exposure-therapy/extraction/extractionScriptTemplate.rxs.Rmd'.
To do the actual extraction, there are two general routes an extractor can take. The first is to use R Studio. The advantage of using R Studio is that, because each extraction script file (rxs file) is in fact an R Markdown file, it can be rendered into a report for the extracted study immediately. This can show whether any mistakes were made during extraction, and easily allows the extractor to check the results of their labour.
However, a disadvantage of R Studio is that R Markdown files are always wrapped. Wrapping means that to prevent the need for horizontal scrolling, long lines of text are displayed on multiple lines. Wrapping is almost always very useful. Text processors, for example, always wrap; text in books is always wrapped; and so is online content.
However, extraction scripts contain very long lines when closely related entities are extracted in list form; in that case, their explanation and examples are placed as comments (preceded by R’s comment symbol, #) behind the entities and values to extract, which can look very confusing if lines are wrapped.
RStudio does use syntax coloring to clearly indicate which parts of the extraction script are comments and which parts are values, but still, extractors might find this confusing.
The second option, therefore, is to use an external editor. For extractors working in a Windows environment, Notepad++ is recommended; for extractors working in a Mac OS environment, BBEdit is recommended (extractors using a Linux distro probably already have their preferred text editors).
Figure 1: Notepad++ when no file has been loaded yet
Working with RStudio requires installing R as well.
When extracting articles, an extractor takes the following steps:
Open the article (this usually means opening the relevant PDF in the pdfs directory).
Copy the extraction script template to a new file in the extraction directory in the study repository.
Give the new file a name conform the following convention: a list of the last names of all authors, all in lower case (i.e. without capitals), separated by dashes (-), and ending with the year of the study, separated from the list of author names by two dashes (--), and ending with the extraction script extension (.rxs.Rmd). Thus, the filename should look something like this: boys-marsden--2003.rxs.Rmd.
Open the new (and newly renamed) extraction script in the editor of choice (see the ‘Software considerations’ section above).
If you haven’t looked at the extraction script yet, study it. If you encounter anything you’re uncertain about, contact another team member to ask them to explain it.
In the extraction script, scroll to the line containing the text START: study (ROOT).
Work your way through the extraction script, completing each applicable extractable entity and removing those that cannot be extracted (see section ‘Extracting entities’ below). Often, the first entity will be the study identifier (usually a Digital Object Identifier or DOI), but completing an extraction script is often a nonlinear activity, starting with the primary entities of interest.
Once you have completed the extraction script, if you use RStudio, you can ‘render’ or ‘knit’ it by clicking the ‘Knit’ button at the top. This will show you what you extracted. If you made any errors (e.g. forgot a comma, or a single or double quote, or forgot to open or close a parenthesis, or mistyped a variable name, etc), this should become clear at this point. Correct any errors. (If you use another editor, you won’t be able to check this at this point.)
Repeat these steps for the next article.
In this process, apply the steps explained in the ‘Extracting entities’ section below. Note that in some cases, you will have to make a choice as to how to extract (code) certain information. The decisions that have already been taken to guide these choices have been listed in the ‘Coding decisions’ section below.
If you run into any problems, clearly write them down, and depending on what you agreed with your team members, accumulate these issues and discuss them at the next meeting, or immediately pass them on using whichever medium you use for coordinating study progress.
The extraction process consists of extracting facts from published reports of empirical research (not always, but almost always). Extraction consists of two steps: studying the report and interpreting its contents to determine what is to be extracted (also called coding) and the actual extraction (documenting the extracted facts). When using metabefor (and we do), facts that can be extracted are called entities. So, can entity is one datum (singular of data) to be extracted, such as a year, a sample size, the measurement level of a variable, a precentage, or an effect size. Extraction scripts (i.e. .rxs.Rmd files) contain a long list of these entities.
Some entities are just single entities. For example, this is an example of the extraction script fragment for extracting a sample size:
############################################################################
############################################################### START: N ###
############################################################################
study$methods$AddChild('N');
study$methods$N[['value']] <-
############################################################################
###
### SAMPLE SIZE
###
### Total number of human participants in the study (note: the
### actual sample size may be larger if multiple observations are
### collected per participant)
###
############################################################################
NA
############################################################################
######################################### VALUE DESCRIPTION AND EXAMPLES ###
############################################################################
###
### Any valid whole number
###
### EXAMPLES:
###
### 30
### 8762
###
############################################################################
study$methods$N[['validation']] <- expression(is.na(VALUE) || (is.numeric(VALUE) && (VALUE%%1==0) && (length(VALUE) == 1)));
############################################################################
################################################################# END: N ###
############################################################################
This excerpt has a number of fragments. As an extractor, not everything is relevant. When extracting, the relevant fragments are the following:
START: N
This signifies that an entity block starts to extract the entity with the identifier (a unique name of this entity) N. For entity blocks that are repeating blocks, this identifier is followed by (REPEATING). Entity blocks that are repeating have to be copied and pasted in the extraction script as often as is required. When copy-pasting, copy-paste the entire entity block.
SAMPLE SIZE
This is the ‘human readable’ (i.e. nicely worded) name for this entity. This will normally just be a clearer version of the identifier.
Total number of human participants in the study (note: the
actual sample size may be larger if multiple observations are
collected per participant)
This is the explanation of what exactly this entity is (i.e. what an extractor should look for in the report, and how to code it).
NA
In between this description of what to code and the following section (see next fragment), the extracted value is entered (this fragment is called the extracted value fragment). In the extraction script template, the default value as specified in the extraction script specification (three spreadsheets that together specify the entities, value templates, and definitions, and that are parsed by metabefor to generate the extraction script template) is prefilled. NA stands for a missing value, in other words, something that has not been extracted yet. Depending on what is specified in the extraction script specification, NA can be a valid value or an invalid value (i.e. if something must be extracted). Another common default value is "" to signify an empty text string.
VALUE DESCRIPTION AND EXAMPLES
This fragment contains a description of the valid values for this entity as well as some examples. This fragment ends with the entity block ending:
END: N
Like for the entity block start, if the entity block is repeating (i.e. describes an entity or entity list that can be extracted multiple times in one report), the identifier (in this example, N) is followed by (REPEATING).
Closely related entities can be combined into an entity list for convenience’s sake. Entity lists look very similar to single entities, but instead of providing a space for extracting a single value in the extracted value fragment, they combine multiple extractable entities in a list (for example, univariate results pertaining to a variable; or associations pertaining to two variables). This list always contains pairs of extractable entities’ names, equals signs (=), and the default value, and these pairs are delimited by comma’s (,). The list of pairs is always preceded by the text list( and followed by a closing parenthesis and a semicolon ();). In entity lists, the explanation of each value as well as the examples are included in comments behind each name-value pair:
##########################################################################
############################### START: operationalisations (REPEATING) ###
##########################################################################
study$methods$variables$AddChild('operationalisations__1__');
study$methods$variables$operationalisations__1__[['value']] <-
##########################################################################
###
### OPERATIONALISATIONS
###
### The operationalisations (measurements or manipulations)
### used in the study
###
##########################################################################
list(oper.name = "Enter name here (mandatory)", ### Variable Name: A unique name (in this study), to be able to refer to this variable later, and to easily find it afterwards [Examples: "Example"; "Another example"] [Value description: A single character value that cannot be omitted]
oper.moment = 1, ### Moment: Moment(s) this variable was measured/manipulated. [Examples: c(23, 62); 52; c(76, 12, 42)] [Value description: A vector of integers (i.e. one or more whole numbers)]
oper.type = NA, ### Type of operationalisation: Whether this variable is a manipulation, a single measurement, or an aggregate [Examples: c("manipulation", "item", "aggregate")] [Value description: A string that has to exactly match one of the values specified in the "values" column of the Coding sheet]
oper.datatype = NA, ### Type of data: "Measurement level" of this operationalisation. [Examples: c("numeric", "logical", "nominal", "ordinal", "string")] [Value description: A string that has to exactly match one of the values specified in the "values" column of the Coding sheet]
oper.values = NA, ### Values: Possible values this variable can take: only valid for "nominal" or "ordinal" variables. [Examples: c(23, 62); 52; c(76, 12, 42)] [Value description: A vector of integers (i.e. one or more whole numbers)]
oper.labels = NA, ### Labels: Labels for the values. [Examples: c("First value", "Second value")] [Value description: A character vector (i.e. one or more strings)]
oper.description = "", ### Description: A description of this variable (can be more extensive than the name) [Examples: "Example"; "Another example"] [Value description: A single character value]
oper.comment = ""); ### Comment: Any relevant comments the coder wants to add [Examples: "Example"; "Another example"] [Value description: A single character value]
##########################################################################
study$methods$variables$operationalisations__1__[['validation']] <- list(`oper.name` = expression(!is.na(VALUE) && !is.null(VALUE) && (nchar(VALUE) > 0)),
`oper.moment` = expression(is.na(VALUE) || (is.numeric(VALUE) && all(VALUE%%1==0))),
`oper.type` = expression(is.na(VALUE) || (VALUE %in% c("manipulation", "item", "aggregate"))),
`oper.datatype` = expression(is.na(VALUE) || (VALUE %in% c("numeric", "logical", "nominal", "ordinal", "string"))),
`oper.values` = expression(is.na(VALUE) || (is.numeric(VALUE) && all(VALUE%%1==0))),
`oper.labels` = expression(is.na(VALUE) || (is.character(VALUE))),
`oper.description` = expression(is.na(VALUE) || (is.character(VALUE) && length(VALUE) == 1)),
`oper.comment` = expression(is.na(VALUE) || (is.character(VALUE) && length(VALUE) == 1)));
study$methods$variables$operationalisations__1__$name <- study$methods$variables$operationalisations__1__$value[['oper.name']];
##########################################################################
################################# END: operationalisations (REPEATING) ###
##########################################################################
Literature syntheses are usually about results from empirical investigations. Therefore, usually univariate results or associations are the primary entity of interest. Therefore, it is usually a good idea to start the extraction with looking for the primary entity (or entities) of interest. For example, if the literature synthesis is about an effect size (i.e. an association), but a report does not provide any information about the relevant association, it may be necessary to contact the authors (see the section ‘Communication with authors’ below). Alternatively, it may be necessary to extract not the association, but univariate results, that will later allow computing the association (see section ‘When associations are not reported’).
If the primary entities of interested are located and extracted, then the included operationalisations (i.e. manifestations of the variables of interest) can be extracted, and in that case, it is also clear that it is worthwhile to extract the methodological details of the study. Thus, in practice, completing an extraction script is a nonlinear process.
Data collection yields two kinds of data: data representing manipulations and data representing measurements.
The values of the first type are determined by the study’s design: for example, in a simple two-cell experiment, such as a randomized controlled trial with two arms, participants in one condition (or ‘arm’) can receive either a placebo (‘control condition’) or an intervention or therapy (‘experimental condition’). In such designs, the condition into which each participant is randomized is usually denoted by a 0 or a 1, which is entered as a variable into the datafile. These values, however, aren’t provided by the participants; they don’t represent collected data, but instead describe the study design, specifically as it pertains to each participant.
The values of the second type, by contrast, usually result from the registration of participants’ responses to stimuli (e.g. participants can endorse certain answer options after having been presented with questions in a questionnaire). For data of this type, realistic ranges for the data will be known in advance (e.g. a questionnaire may register responses on a 5-point scale), but the distributions of the data are unknown (e.g. means, standard deviations, distribution shapes, etc).
The goal of empirical research in psychology is often to estimate the association between two (or sometimes more) variables (if a research question concerns causality, at least one variable is operationalised as a manipulation; otherwise, both may be operationalised as measurements).
Datasets (sets of one or more data series, where a data series is a set of one or more data points, where a data point is one number that represents a value of an operationalisation, e.g. a 0 or a 1 representing a manipulation, or a 4 representing the registered reponse in a questionnaire) can be subjected to many different analyses. In the context of research syntheses such as systematic reviews and meta-analyses, it is useful to distinguish three types of analyses: univariate analyses, bivariate analyses, and multivariate analyses.
Univariate analyses are analyses conducted on one data series. This can be a data series as collected (e.g. the responses registered from participants for one item in a questionnaire) or a data series that represents an aggregation of multiple data series (e.g. a mean of the data series that each represent participants’ answers to a different item in a questionnaire). Univariate analyses by definition do not provide information about associations between variables, but only about the distributions of data points. Examples of univariate results are:
Like any number estimated from a sample, the point estimate for each of these results will vary from sample to sample due to sampling and measurement error. Therefore, each univariate result may be reported with confidence intervals as well.
Univariate results are normally not reported for data series that represent manipulations.
Bivariate analyses are analyses conducted on two data series. Again, the data series may be aggregates of other data series, and if a design is experimental, the main research question is usually answered by an analysis that includes a data series representing a manipulation.
In the frequentist tradition, there are two types of bivariate analyses: tests and effect sizes. Tests are used in a null hypothesis significance testing framework, and were traditionally means to arrive at p-values. Test statistics and p-values provide no information about the strength of an association unless the corresponding degrees of freedom are taken into account. Effect sizes do the opposite: they provide information as to the strength of an association. If both effect sizes and test statistics are available, effect sizes should be extracted instead of test statistics (unless both are extracted, which would then allow rudimentary verification of the reported results).
There are a few common bivariate test statiatics:
There are also a few common bivariate effect sizes:
Sometimes researchers report multivariate results: results based on more than two data series. In the context of research syntheses, this can be challenging: estimates from multivariate analyses are conditional upon the entire model (i.e. all data series in the model). In other words, it is not always possible (and will often be problematic) to meta-analytically aggregate effect sizes from bivariate analyses and from multivariate analyses; or from multivariate analyses alone, unless those analyses were the same.
There are a few common multivariate analyses:
If in a multivariate analyses, the predictors are all independent (orthogonal, not associated to each other), for example if the factors in a factorial anova represent conditions in an experiment with equal cell sizes, and not covariates were included, the results can be aggregated without any problems.
When associations are not reported, look for univariate data, such as means and standard deviations. In an experimental design with multiple measurement moments, each of these extracted univariate results must have a specified measurement moment and a specified value for the subsample. The subsamples, then are defined as arms of the experiment (e.g. values of the manipulated variable). This requires defining a subsample for every arm, as well as variables (or more accurately, operationalisations) for the manipulation (or treatment, or intervention, etc) and the dependent variable(s). In addition, the values of these variables (i.e. the treatment, which will be, for example, 0 or 1 to denote absence or presence, and a dependent variable, which can be, for example, a mean and a standard deviation). Therefore, in this case, you need to extract five entity lists to store one set of univariate results. To illustrate this, an example of each of these five entity lists (in these examples, only the extracted value fragments are shown):
list(subsample.name = "waitinglist",
subsample.N = 14,
subsample.age = NA);
This entity list (with three extracted entities) specifies a subsample. This subsample codes for one group in the experiment: we will specify which group below. Splitting the sample into subsamples enables specifying, for example, the mean for this subsample.
list(oper.name = "treatment",
oper.moment = c(1, 2),
oper.type = "manipulation",
oper.datatype = "logical",
oper.values = c(0, 1),
oper.labels = c("Control", "Experimental"),
oper.description = "The treatment arms",
oper.comment = "");
This entity list specifies an operationalisation. This operationalisation is the manipulation in an experimental design (as can be seen from oper.type), with three possible conditions (Control and Experimental). This operationalisation allows us to specify that the waitinglist subsample is the Control group in this study. Extracted entity oper.moment shows that this study has (at least) two measurement moments, and that this operationalisation is the same at both moments (in other words, given that this concerns the manipulation, this simply means that participants, once randomized into their condition, stay in that condition at the second measurement moment).
list(oper.name = "SCARED",
oper.moment = c(1, 2),
oper.type = "aggregate",
oper.datatype = "numeric",
oper.description = "The SCARED questionnaire",
oper.comment = "");
This second operationalisation specifies a dependent variable called SCARED, a measurement instrument for symptoms of anxiety disorders. This is the variable of which we will want to extract the mean and standard deviation for the control condition. This operationalisation, too, is the same at both measurement moments (oper.moment = c(1, 2) means that this information pertains to both measurement moments).
list(uni.name = "treatment_t1_control",
uni.variable = "treatment",
uni.subsample = "waitinglist",
uni.moment = c(1, 2),
uni.category = 0);
This univariate result codes the condition of this subsample: variable treatment (as specified in the relevant operationalisation entity list) has value 0 at moments 1 and 2 for subsample waitinglist. Combined with the operationalisation specified earlier, this means that this subsample is in the control condition.
list(uni.name = "symptoms_t1_control",
uni.variable = "SCARED",
uni.subsample = "waitinglist",
uni.moment = 1,
uni.mean = 26.93,
uni.sd = 4.56,
uni.comment = "Extracted from Table 2, page 71");
Finally, now that subsample has been defined; both variables have been defined; and the condition of this subsample has been defined; the actual univariate results of interest can be coded. In this case, this concerns a mean and a standard deviation. Two means, standard deviations, and the corresponding sample sizes allow computing Cohen’s d, an effect size measure. metabefor will handle these conversions.
When outcomes for multiple psychopathologies are reported, only fear or anxiety is extracted.
When multiple outcome measures are reported, self-reported symptoms (or reduction in reported symptoms) is extracted.
When multiple self-reported symptoms are reported, both for anxiety in general and for one or more specific anxiety disorders, anxiety in general is extracted.
When multiple outcome measures are reported, the measure that measures fear or anxiety at the most generic level is extracted (e.g. measures of specific anxiety disorderd are avoided to the benefit of generic measures whenever possible).
When multiple measurement moments were reported, the measurement moment immediately following the conclusion of the therapy sessions was extracted.
When coding the therapy ingredients, because authors generally do not list everything that was not done in a therapy protocol, absence is generally coded as such based on how authors describe the therapy. For example, absence of an ingredient can be sensibly inferred if presence of that ingredient would be incompatible with other ingredients that were present, or with the rationale of the therapy as described in the manual. In the rare cases where absence of an ingredient is explicitly stated in the therapy description or manual, this will be indicated in the study description.
This table shows the list of therapy ingredients to be coded, for easy access and scrutiny, since it’s so central to this study.
| Name | Description | |
|---|---|---|
| treatments | Treatments | The treatments for each group |
| treatment | Treatment | Details about the delivered treatment for this group (this operationalisation value) |
| treatment.name | Treatment Name | A unique name (identifier) for this treatment. |
| treatment.variable | Treatment Variable | Name of the treatment variable. |
| treatment.sessions | Number of sessions | The number of sessions in this therapy group (condition or arm). |
| treatment.operValue | Group | Which treatment group (operationalisation value of the treatment variable) this description pertains to. |
| treatment.manualAvailable | Treatment manual available | Whether a treatment manual was available for coding. If not, the ingredients were coded based on the article. |
| fearHierarchy | Fear hierarchy | The exposure trials are rank-ordered in their ability to elicit anxiety. |
| fearMon | Fear monitoring | The fear level of the participant is being monitored throughout the exposure trial. |
| lvlBasedFear | Level-based termination (fear) | The end of the exposure trial is determined by fear reduction to a certain level, as measured through monitoring (e.g. if this therapy ingredient is coded, ‘Fear monitoring’ must always also be coded). |
| lvlBasedExpect | Level-based termination (expectancy) | The end of the exposure trial is determined by prespecified criteria of expectancy disconfirmation, as tested during the exposure trial (e.g. if this therapy ingredient is coded, ‘Expectancy monitoring’ must always also be coded). |
| expectState | Expectancy statement | Expectancies for aversive events are explicitly stated before the start of the exposure trial. |
| expectMonitor | Expectancy monitoring | Expectancy disconfirmation for aversive events are monitored throughout the exposure trial. |
| learnEval | Learning evaluation | Following the exposure trial, the participants are asked to evaluate what they learned regarding the non-occurrence of the feared event, discrepancies between what was predicted and what occurred. |
| expBasedFear | Exposure-based termination (fear) | The exposure trial ends when prespecified conditions that are based on expected effects of the exposure on fear are met (e.g. exposure for a specific period). |
| expBasedExpect | Exposure-based termination (expectancy) | The exposure trial ends when prespecified conditions that are based on expected effects of the exposure on expectancies are met (e.g. exposure for a specific period). |
| expBasedNonspec | Exposure-based termination (nonspecific) | The exposure trial ends when prespecified conditions are met (e.g. exposure for a specific period), where it is not specified what these conditions are based on. |
| lessOverest | Lessen overestimation | Cognitive interventions designed to lessen probability overestimation (e.g., “I am unlikely to be bitten by the dog”) and perceived negative valence (e.g., “It is not so bad to be rejected”) occur. |
| deepExt | Deepened extinction | Combination of multiple cues (internal and/or external) during exposure therapy, after initially conducting some exposure to each cue in isolation. |
| pharma | Pharmacological enhancement | Pharmacological enhancement: Pharmacological means are used to support the therapy. |
| occReinfExt | Occasional reinforced extinction | Occasional CS-US pairings during extinction training occur. |
| removSaf | Removal of safety signals | The prevention or removal of “safety signals” or “safety behaviors” during the exposure therapy occurs. |
| variab | Variability | Exposure is conducted with varying stimuli, for varying durations, at varying levels of intensity, or items from a fear hierarchy are selected out of order, rather than continuing exposure in one situation until fear declines before moving to the next situation. |
| retrCues | Retrieval cues | Retrieval cues (of the CS-no US association) are included during extinction training to be used in other contexts once extinction is over. |
| multiCont | Multiple contexts | Interoceptive, imaginal, and in vivo exposures are conducted in multiple different contexts, such as when alone, in unfamiliar places, or at varying times of day or varying days of the week. |
| preExposure | Brief pre-exposure | The phobic stimulus is introduced for a brief period about 30 minutes before repeated trials of exposure. |
| contingency | Contingency management | Any use of rewards on the basis of progress in the exposure therapy. |
| expNature | Nature of exposure | The type of exposure used (in vivo, in vitro, or virtual reality). |
Warning: package 'pander' was built under R version 3.5.1
| treatment.name | treatment.variable | treatment.sessions | treatment.operValue | treatment.manualAvailable | fearHierarchy | fearMon | lvlBasedFear | lvlBasedExpect | expectState | expectMonitor | learnEval | expBasedFear | expBasedExpect | expBasedNonspec | lessOverest | deepExt | pharma | occReinfExt | removSaf | variab | retrCues | multiCont | preExposure | contingency | expNature | study | X.in.vitro. |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| GBAT | treatment | 10 | 1 | yes | present | absent | absent | absent | present | absent | present | absent | absent | present | absent | absent | absent | absent | absent | absent | absent | absent | absent | present | in vivo | chu-crocco-esseling-areizaga-lindner-skriner–2015.rxs.Rmd | NA |
| CBT modular | treatment | 8 | 1 | no | present | unknown | unknown | unknown | unknown | unknown | unknown | unknown | unknown | unknown | unknown | unknown | unknown | unknown | unknown | unknown | unknown | unknown | unknown | unknown | unknown | ginsburg-becker-drazdowski-tein–2011.rxs.Rmd | NA |
| CBT Silverman | treatment | 10 | 1 | no | present | unknown | unknown | unknown | unknown | unknown | unknown | unknown | unknown | unknown | unknown | unknown | unknown | unknown | unknown | unknown | unknown | unknown | unknown | present | in vivo | ginsburg-drake–2002.rxs.Rmd | NA |
| Coping Cat | treatment | 16 | 1 | yes | present | present | present | absent | present | absent | absent | present | absent | absent | present | absent | absent | absent | present | absent | absent | present | absent | present | in vivo | kendall-flannery-panichelli-southam-henin-warman–1997.rxs.Rmd | in vitro |
| CBT ??st | enter < |
1 | 1 | yes | present | present | present | absent | absent | absent | unknown | absent | absent | absent | unknown | absent | absent | unknown | unknown | absent | unknown | absent | unknown | absent | in vivo | leutgeb-schafer-kochel-schienle–2012.rxs.Rmd | NA |
| CBT trauma | treatment | 9 | 1 | yes | present | present | present | absent | present | absent | absent | absent | absent | present | present | absent | absent | absent | present | absent | unknown | present | absent | present | in vivo | nixon-sterk-pearce–2012.rxs.Rmd | NA |
| CBT trauma | treatment | 9 | 1 | yes | present | present | present | absent | present | absent | absent | absent | absent | present | present | absent | absent | absent | present | absent | unknown | present | absent | present | in vitro | nixon-sterk-pearce–2012.rxs.Rmd | NA |
| NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | peris-kendall-sherrill-gosh-rhyn—2015.rxs.Rmd | NA |
| PLET | treatment | 10 | 1 | no | unknown | unknown | unknown | unknown | unknown | unknown | unknown | unknown | unknown | unknown | unknown | unknown | unknown | unknown | unknown | unknown | unknown | unknown | unknown | unknown | unknown | rudy-zavrou-johnco-storch-lewin–2017.rxs.Rmd | NA |
| CBT Spence | treatment | 12 | 1 | no | present | unknown | unknown | unknown | unknown | unknown | unknown | unknown | unknown | unknown | unknown | unknown | unknown | unknown | unknown | unknown | unknown | unknown | unknown | unknown | in vivo | spence-donovan-brechman–2000.rxs.Rmd | NA |
| CBT CGC/ARC | treatment | 16 | 1 | no | present | present | present | absent | absent | absent | absent | absent | absent | absent | present | absent | absent | absent | absent | unknown | absent | present | absent | absent | in vivo | sung-ooi-goh-pathy-fung-ang-chua-lam–2011.rxs.Rmd | NA |
| CBT CGC/ARC | treatment | 16 | 1 | no | present | present | present | absent | absent | absent | absent | absent | absent | absent | present | absent | absent | absent | absent | unknown | absent | present | absent | absent | in vitro | sung-ooi-goh-pathy-fung-ang-chua-lam–2011.rxs.Rmd | NA |
| PC-Exp | treatment | 6 | 1 | yes | present | present | absent | absent | present | present | present | absent | absent | present | present | unknown | absent | absent | present | absent | unknown | unknown | absent | present | in vivo | whiteside-ale-young-dammann-tiede-biggs–2015.rxs.Rmd | NA |
| BIACA | treatment | 16 | 1 | yes | present | present | absent | absent | present | unknown | present | present | absent | absent | present | absent | absent | absent | present | absent | absent | present | absent | present | in vivo | wood-ehrenreich-alessandri-fujii-renno-laugeson-piacentini-denadai-arnold-lewin-murphy-stoch–2015.rxs.Rmd | NA |
| BIACA | treatment | 16 | 1 | yes | present | present | absent | absent | present | unknown | present | present | absent | absent | present | absent | absent | absent | present | absent | absent | present | absent | present | in vitro | wood-ehrenreich-alessandri-fujii-renno-laugeson-piacentini-denadai-arnold-lewin-murphy-stoch–2015.rxs.Rmd | NA |
Warning: attributes are not identical across measure variables;
they will be dropped
| study | treatmentName | manualAvailable | age | anxiety | N | country |
|---|---|---|---|---|---|---|
| chu et al. (2015) | GBAT | yes | 12-14 | generalized anxiety, social anxiety & separation anxiety | 35 | US |
| ginsburg et al. (2011) | CBT modular | no | 7-17 | generalized anxiety, social anxiety, separation anxiety, specific phobia & unspecified anxiety | 32 | US |
| ginsburg et al. (2002) | CBT Silverman | no | 14-17 | generalized anxiety, specific fobia, agoraphobia & social anxiety | 12 | US |
| kendall et al. (1997) | Coping Cat | yes | 9-13 | generalized anxiety, social anxiety & separation anxiety | 94 | US |
| leutgeb et al. (2012) | CBT ??st | yes | 8-13 | specific phobia | 32 | AT |
| nixon et al. (2012) | CBT trauma | yes | 7-17 | posttraumatic stress disorder | 33 | AU |
| peris et al. (2015) | NA | NA | 7-17 | separation anxiety, social anxiety & generalized anxiety | 488 | US |
| rudy et al. (2017) | PLET | no | 4-7 | generalized anxiety, social anxiety, separation anxiety, specific phobia, obsessive compulsive disorder & other | 22 | US |
| spence et al. (2000) | CBT Spence | no | 7-14 | social anxiety | 50 | AU |
| sung et al. (2011) | CBT CGC/ARC | no | 9-16 | generalized anxiety, social anxiety, separation anxiety, obsessive compulsive disorder, panic disorder & other | 70 | SG |
| whiteside et al. (2015) | PC-Exp | yes | 7-14 | generalized anxiety, social anxiety & separation anxiety | 14 | US |
| wood et al. (2015) | BIACA | yes | 11-15 | separation anxiety, social anxiety, generalized anxiety & obsessive compulsive disorder | 33 | US |
comments
Ik ben geen expert op dit gebied, maar ik hoop voor je analyse dat uitkomstmaten te vergelijken zijn voor beide theorieeen. Habituatie is veelal ouder, dus daar worden misschien niet alleen andere proximale maten gebruikt, maar ook bijv. DSM criteria komen uit een eerdere DSM versie dan de meer recente studies over inhibitie.
Presumably you have already gone through the papers by Michelle Craske on maximising exposure therapy.
No
NA
NA